Perspectives bayésiennes en géographie

Abstract

Bayesian Perspectives in Geography. — The purpose of this article is to show the relevance of the subjective approach to probability for geographical research. Consistent with a certain epistemological trend in the human sciences, it provides geographers with a simple method of analysis. So-called Bayesian analysis allows an individual (or a group) who has to make a decision under uncertainty to choose a strategy which is consistent with his own preferences and personal judgments. It is first explained in function of its main characteristics : combination of the estimation of the uncertain phenomena with the preferences for the various consequences; and sequential decision process. The mathematical technique of the analysis is then clarified by means of a numerical example, and the few norms required are mentioned. Then, its applicability to geography is discussed and demonstrated in relation to perception, limited knowledge of the environment, level of satisfaction, learning and decision-making.